backedup 5.0.0.beta.3

data/lib/backup/storage/swift.rb

@@ -0,0 +1,124 @@
+require "backup/cloud_io/swift"
+
+module Backup
+  module Storage
+    class Swift < Base
+      include Storage::Cycler
+      class Error < Backup::Error; end
+
+      ##
+      # Swift credentials
+      attr_accessor :username, :password
+
+      ##
+      # Keystone tenant name if using v2 auth
+      attr_accessor :tenant_name
+
+      ##
+      # Swift container name
+      attr_accessor :container
+
+      ##
+      # Swift region. It might be required for certain providers
+      attr_accessor :region
+
+      ##
+      # OpenStack keystone url
+      attr_accessor :auth_url
+
+      ##
+      # Number of times to retry failed operations.
+      #
+      # Default: 10
+      attr_accessor :max_retries
+
+      ##
+      # Time in seconds to pause before each retry.
+      #
+      # Default: 30
+      attr_accessor :retry_waitsec
+
+      ##
+      # The size of the batch operations (delete/list/etc.) in your
+      # OpenStack deployment
+      #
+      # Default: 1000
+      attr_accessor :batch_size
+
+      ##
+      # Additional options to pass along to fog.
+      # e.g. Fog::Storage.new({ :provider => 'Swift' }.merge(fog_options))
+      attr_accessor :fog_options
+
+      def initialize(mode, storage_id = nil)
+        super
+
+        @max_retries    ||= 10
+        @retry_waitsec  ||= 30
+        @path           ||= "backups"
+        @batch_size     ||= 1000
+        @fog_options    ||= {}
+
+        @path = @path.sub(%r{^\/}, "")
+
+        check_configuration
+      end
+
+      private
+
+      def cloud_io
+        @cloud_io ||= CloudIO::Swift.new(
+          username:             username,
+          password:             password,
+          tenant_name:          tenant_name,
+          region:               region,
+          container:            container,
+          auth_url:             auth_url,
+          max_retries:          max_retries,
+          retry_waitsec:        retry_waitsec,
+          batch_size:           batch_size,
+          fog_options:          fog_options
+        )
+      end
+
+      def transfer!
+        package.filenames.each do |filename|
+          src = File.join(Config.tmp_path, filename)
+          dest = File.join(remote_path, filename)
+          Logger.info "Storing '#{container}/#{dest}'..."
+          cloud_io.upload(src, dest)
+        end
+      end
+
+      # Called by the Cycler.
+      # Any error raised will be logged as a warning.
+      def remove!(package)
+        Logger.info "Removing backup package dated #{package.time}..."
+
+        remote_path = remote_path_for(package)
+        objects = cloud_io.objects(remote_path)
+
+        raise Error, "Package at '#{remote_path}' not found" if objects.empty?
+
+        cloud_io.delete(objects)
+      end
+
+      def check_configuration
+        if auth_url.nil?
+          raise Error, <<-EOS
+            Configuration Error
+            Swift auth_url is required
+          EOS
+        end
+
+        required = [:username, :password, :container, :auth_url]
+        required << :tenant_name if auth_url =~ /v2/
+
+        raise Error, <<-EOS if required.map { |name| send(name) }.any?(&:nil?)
+          Configuration Error
+          #{required.map { |name| "##{name}" }.join(", ")} are all required
+        EOS
+      end
+    end
+  end
+end

data/lib/backup/storage/webdav.rb

@@ -0,0 +1,102 @@
+# encoding: utf-8
+require 'faraday'
+
+module Backup
+  module Storage
+    class Webdav < Base
+      include Storage::Cycler
+
+      ##
+      # Webdav credentials
+      attr_accessor :username, :password
+
+      ##
+      # Server hostname and port, SSL settings
+      attr_accessor :ip, :port, :use_ssl, :ssl_verify
+
+      ##
+      # configure a connection timeout
+      attr_accessor :timeout
+
+      def initialize(model, storage_id = nil)
+        super
+
+        @port       ||= 80
+        @use_ssl    ||= false
+        @ssl_verify = true if @ssl_verify.nil?
+        @path       ||= 'backups'
+        @timeout    ||= nil
+        path.sub!(/^~\//, '')
+      end
+
+      private
+
+      ##
+      # create the connection object used for initializing the requests
+      #
+      # Note:
+      # Webdav has a special HTTP verb for creating collections that needs to
+      # be added the the list of allowed verbs in Faraday
+      def connection
+        if Faraday::Connection.const_defined?(:METHODS)
+          Faraday::Connection::METHODS.add(:mkcol)
+        end
+
+        conn_hash = {
+          url: base_url,
+          request: {
+            timeout: timeout,
+            open_timeout: timeout
+          },
+          ssl: { verify: ssl_verify }
+        }
+
+        Faraday::Connection.new(conn_hash) do |builder|
+          builder.request :multipart
+          builder.request :basic_auth, username, password
+          builder.adapter :net_http
+        end
+      end
+
+      def base_url
+        "#{(use_ssl ? 'https' : 'http')}://#{ip}:#{port}"
+      end
+
+      def transfer!
+        create_remote_path
+
+        package.filenames.each do |filename|
+          src = File.join(Config.tmp_path, filename)
+          dest = File.join(remote_path, filename)
+
+          Logger.info "Storing '#{ ip }:#{ dest }'..."
+          connection.put(dest) do |req|
+            req.headers['Content-Type'] = 'octet/stream'
+            req.headers['Transfer-Encoding'] = 'chunked'
+            req.headers['Content-Length'] = File.size(src).to_s
+            req.body = Faraday::UploadIO.new(src, 'octet/stream')
+          end
+        end
+      end
+
+      def create_remote_path
+        path_parts = Array.new
+        remote_path.split('/').each do |path_part|
+          path_parts << path_part
+          connection.run_request(:mkcol, path_parts.join('/'), nil, nil)
+        end
+      end
+
+      ##
+      # called by the Cycler.
+      def remove!(package)
+        Logger.info "Removing backup package dated #{ package.time }..."
+        remote_path = remote_path_for(package)
+        package.filenames.each do |filename|
+          connection.delete(File.join(remote_path, filename))
+        end
+        connection.delete(File.join(remote_path, ''))
+      end
+    end
+  end
+end

data/lib/backup/syncer/base.rb

@@ -0,0 +1,67 @@
+module Backup
+  module Syncer
+    class Base
+      include Utilities::Helpers
+      include Config::Helpers
+
+      ##
+      # Path to store the synced files/directories to
+      attr_accessor :path
+
+      ##
+      # Flag for mirroring the files/directories
+      attr_accessor :mirror
+
+      ##
+      # Optional user-defined identifier to differentiate multiple syncers
+      # defined within a single backup model. Currently this is only used
+      # in the log messages.
+      attr_reader :syncer_id
+
+      attr_reader :excludes
+
+      def initialize(syncer_id = nil)
+        @syncer_id = syncer_id
+
+        load_defaults!
+
+        @mirror ||= false
+        @directories ||= []
+        @excludes ||= []
+      end
+
+      ##
+      # Syntactical suger for the DSL for adding directories
+      def directories(&block)
+        return @directories unless block_given?
+        instance_eval(&block)
+      end
+
+      def add(path)
+        directories << path
+      end
+
+      # For Cloud Syncers, +pattern+ can be a string (with shell-style
+      # wildcards) or a regex.
+      # For RSync, each +pattern+ will be passed to rsync's --exclude option.
+      def exclude(pattern)
+        excludes << pattern
+      end
+
+      private
+
+      def syncer_name
+        @syncer_name ||= self.class.to_s.sub("Backup::", "") +
+          (syncer_id ? " (#{syncer_id})" : "")
+      end
+
+      def log!(action)
+        msg = case action
+              when :started then "Started..."
+              when :finished then "Finished!"
+              end
+        Logger.info "#{syncer_name} #{msg}"
+      end
+    end
+  end
+end

data/lib/backup/syncer/cloud/base.rb

@@ -0,0 +1,176 @@
+module Backup
+  module Syncer
+    module Cloud
+      class Error < Backup::Error; end
+
+      class Base < Syncer::Base
+        MUTEX = Mutex.new
+
+        ##
+        # Number of threads to use for concurrency.
+        #
+        # Default: 0 (no concurrency)
+        attr_accessor :thread_count
+
+        ##
+        # Number of times to retry failed operations.
+        #
+        # Default: 10
+        attr_accessor :max_retries
+
+        ##
+        # Time in seconds to pause before each retry.
+        #
+        # Default: 30
+        attr_accessor :retry_waitsec
+
+        def initialize(syncer_id = nil, &block)
+          super
+          instance_eval(&block) if block_given?
+
+          @thread_count   ||= 0
+          @max_retries    ||= 10
+          @retry_waitsec  ||= 30
+
+          @path ||= "backups"
+          @path = path.sub(/^\//, "")
+        end
+
+        def perform!
+          log!(:started)
+          @transfer_count = 0
+          @unchanged_count = 0
+          @skipped_count = 0
+          @orphans = thread_count > 0 ? Queue.new : []
+
+          directories.each { |dir| sync_directory(dir) }
+          orphans_result = process_orphans
+
+          Logger.info "\nSummary:"
+          Logger.info "\s\sTransferred Files: #{@transfer_count}"
+          Logger.info "\s\s#{orphans_result}"
+          Logger.info "\s\sUnchanged Files: #{@unchanged_count}"
+          Logger.warn "\s\sSkipped Files: #{@skipped_count}" if @skipped_count > 0
+          log!(:finished)
+        end
+
+        private
+
+        def sync_directory(dir)
+          remote_base = path.empty? ? File.basename(dir) :
+                                      File.join(path, File.basename(dir))
+          Logger.info "Gathering remote data for '#{remote_base}'..."
+          remote_files = get_remote_files(remote_base)
+
+          Logger.info("Gathering local data for '#{File.expand_path(dir)}'...")
+          local_files = LocalFile.find(dir, excludes)
+
+          relative_paths = (local_files.keys | remote_files.keys).sort
+          if relative_paths.empty?
+            Logger.info "No local or remote files found"
+          else
+            Logger.info "Syncing..."
+            sync_block = proc do |relative_path|
+              local_file  = local_files[relative_path]
+              remote_md5  = remote_files[relative_path]
+              remote_path = File.join(remote_base, relative_path)
+              sync_file(local_file, remote_path, remote_md5)
+            end
+
+            if thread_count > 0
+              sync_in_threads(relative_paths, sync_block)
+            else
+              relative_paths.each(&sync_block)
+            end
+          end
+        end
+
+        def sync_in_threads(relative_paths, sync_block)
+          queue = Queue.new
+          queue << relative_paths.shift until relative_paths.empty?
+          num_threads = [thread_count, queue.size].min
+          Logger.info "\s\sUsing #{num_threads} Threads"
+          threads = Array.new(num_threads) do
+            Thread.new do
+              loop do
+                path = queue.shift(true) rescue nil
+                path ? sync_block.call(path) : break
+              end
+            end
+          end
+
+          # abort if any thread raises an exception
+          while threads.any?(&:alive?)
+            if threads.any? { |thr| thr.status.nil? }
+              threads.each(&:kill)
+              Thread.pass while threads.any?(&:alive?)
+              break
+            end
+            sleep num_threads * 0.1
+          end
+          threads.each(&:join)
+        end
+
+        # If an exception is raised in multiple threads, only the exception
+        # raised in the first thread that Thread#join is called on will be
+        # handled. So all exceptions are logged first with their details,
+        # then a generic exception is raised.
+        def sync_file(local_file, remote_path, remote_md5)
+          if local_file && File.exist?(local_file.path)
+            if local_file.md5 == remote_md5
+              MUTEX.synchronize { @unchanged_count += 1 }
+            else
+              Logger.info("\s\s[transferring] '#{remote_path}'")
+              begin
+                cloud_io.upload(local_file.path, remote_path)
+                MUTEX.synchronize { @transfer_count += 1 }
+              rescue CloudIO::FileSizeError => err
+                MUTEX.synchronize { @skipped_count += 1 }
+                Logger.warn Error.wrap(err, "Skipping '#{remote_path}'")
+              rescue => err
+                Logger.error(err)
+                raise Error, <<-EOS
+                  Syncer Failed!
+                  See the Retry [info] and [error] messages (if any)
+                  for details on each failed operation.
+                EOS
+              end
+            end
+          elsif remote_md5
+            @orphans << remote_path
+          end
+        end
+
+        def process_orphans
+          if @orphans.empty?
+            return mirror ? "Deleted Files: 0" : "Orphaned Files: 0"
+          end
+
+          if @orphans.is_a?(Queue)
+            @orphans = Array.new(@orphans.size) { @orphans.shift }
+          end
+
+          if mirror
+            Logger.info @orphans.map { |path|
+              "\s\s[removing] '#{path}'"
+            }.join("\n")
+
+            begin
+              cloud_io.delete(@orphans)
+              "Deleted Files: #{@orphans.count}"
+            rescue => err
+              Logger.warn Error.wrap(err, "Delete Operation Failed")
+              "Attempted to Delete: #{@orphans.count} " \
+              "(See log messages for actual results)"
+            end
+          else
+            Logger.info @orphans.map { |path|
+              "\s\s[orphaned] '#{path}'"
+            }.join("\n")
+            "Orphaned Files: #{@orphans.count}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/backup/syncer/cloud/cloud_files.rb

@@ -0,0 +1,81 @@
+require "backup/cloud_io/cloud_files"
+
+module Backup
+  module Syncer
+    module Cloud
+      class CloudFiles < Base
+        class Error < Backup::Error; end
+
+        ##
+        # Rackspace CloudFiles Credentials
+        attr_accessor :username, :api_key
+
+        ##
+        # Rackspace CloudFiles Container
+        attr_accessor :container
+
+        ##
+        # Rackspace AuthURL (optional)
+        attr_accessor :auth_url
+
+        ##
+        # Rackspace Region (optional)
+        attr_accessor :region
+
+        ##
+        # Rackspace Service Net
+        # (LAN-based transfers to avoid charges and improve performance)
+        attr_accessor :servicenet
+
+        ##
+        # Additional options to pass along to fog.
+        # e.g. Fog::Storage.new({ :provider => 'Rackspace' }.merge(fog_options))
+        attr_accessor :fog_options
+
+        def initialize(syncer_id = nil)
+          super
+
+          @servicenet ||= false
+
+          check_configuration
+        end
+
+        private
+
+        def cloud_io
+          @cloud_io ||= CloudIO::CloudFiles.new(
+            username: username,
+            api_key: api_key,
+            auth_url: auth_url,
+            region: region,
+            servicenet: servicenet,
+            container: container,
+            max_retries: max_retries,
+            retry_waitsec: retry_waitsec,
+            # Syncer can not use SLOs.
+            segments_container: nil,
+            segment_size: 0,
+            fog_options: fog_options
+          )
+        end
+
+        def get_remote_files(remote_base)
+          hash = {}
+          cloud_io.objects(remote_base).each do |object|
+            relative_path = object.name.sub(remote_base + "/", "")
+            hash[relative_path] = object.hash
+          end
+          hash
+        end
+
+        def check_configuration
+          required = %w[username api_key container]
+          raise Error, <<-EOS if required.map { |name| send(name) }.any?(&:nil?)
+            Configuration Error
+            #{required.map { |name| "##{name}" }.join(", ")} are all required
+          EOS
+        end
+      end # class Cloudfiles < Base
+    end # module Cloud
+  end
+end

data/lib/backup/syncer/cloud/local_file.rb

@@ -0,0 +1,97 @@
+require "digest/md5"
+
+module Backup
+  module Syncer
+    module Cloud
+      class LocalFile
+        attr_reader :path
+        attr_accessor :md5
+
+        class << self
+          # Returns a Hash of LocalFile objects for each file within +dir+,
+          # except those matching any of the +excludes+.
+          # Hash keys are the file's path relative to +dir+.
+          def find(dir, excludes = [])
+            dir = File.expand_path(dir)
+            hash = {}
+            find_md5(dir, excludes).each do |file|
+              hash[file.path.sub(dir + "/", "")] = file
+            end
+            hash
+          end
+
+          # Return a new LocalFile object if it's valid.
+          # Otherwise, log a warning and return nil.
+          def new(*args)
+            file = super
+            if file.invalid?
+              Logger.warn("\s\s[skipping] #{file.path}\n" \
+                          "\s\sPath Contains Invalid UTF-8 byte sequences")
+              file = nil
+            end
+            file
+          end
+
+          private
+
+          # Returns an Array of file paths and their md5 hashes.
+          def find_md5(dir, excludes)
+            found = []
+            (Dir.entries(dir) - %w[. ..]).map { |e| File.join(dir, e) }.each do |path|
+              if File.directory?(path)
+                unless exclude?(excludes, path)
+                  found += find_md5(path, excludes)
+                end
+              elsif File.file?(path)
+                if file = new(path)
+                  unless exclude?(excludes, file.path)
+                    file.md5 = Digest::MD5.file(file.path).hexdigest
+                    found << file
+                  end
+                end
+              end
+            end
+            found
+          end
+
+          # Returns true if +path+ matches any of the +excludes+.
+          # Note this can not be called if +path+ includes invalid UTF-8.
+          def exclude?(excludes, path)
+            excludes.any? do |ex|
+              if ex.is_a?(String)
+                File.fnmatch?(ex, path)
+              elsif ex.is_a?(Regexp)
+                ex.match(path)
+              end
+            end
+          end
+        end
+
+        # If +path+ contains invalid UTF-8, it will be sanitized
+        # and the LocalFile object will be flagged as invalid.
+        # This is done so @file.path may be logged.
+        def initialize(path)
+          @path = sanitize(path)
+        end
+
+        def invalid?
+          !!@invalid
+        end
+
+        private
+
+        def sanitize(str)
+          str.each_char.map do |char|
+            begin
+              char.unpack("U")
+              char
+            rescue
+              @invalid = true
+              "\xEF\xBF\xBD" # => "\uFFFD"
+            end
+          end.join
+        end
+      end
+    end
+  end
+end